Upload
edison-hurtado
View
212
Download
0
Embed Size (px)
Citation preview
7/29/2019 2012 Goldthorpe Sociologia de La Desigualdad
1/15
John H. Goldthorpe: Nuffield College | [email protected]
Reis 137, enero-marzo 2012, pp. 201-216
Back to Class and Status: Or Why aSociological View of Social Inequality Should
Be ReassertedDe vuelta a la clase y el estatus: por qu debe revindicarse una perspectiva
sociolgica de la desigualdad social
John H. Goldthorpe
doi:10.5477/cis/reis.137.201
INTRODUCTION
During the period of the long boom, fromthe end of the Second World War through tothe 1970s, it was widely believed that socialinequalities, of both condition and opportuni-ty, were in long-term decline. From the side
of economics, the Kuznets curve was takento show that income inequality widened inthe take-off into industrialism but then na-rrowed as economic growth continued (Kuz-nets, 1955). And it further appeared that, inmore advanced societies, the transmission ofeconomic advantage and disadvantage from
Key words
Social inequality
Social class Socialstatus Social mobility Economists Epide-miologists
AbstractOf late, issues of social inequality have assumed a new political centrality
in many western societies. However, in much discussion of these issues,sociological approaches to the analysis of social inequality have beendisregarded, especially in the work of economists and epidemiologists.The main features of the sociological approach are the emphasis givento inequality in a relational rather than a merely attributional sense, and tothe distinction between social class and social status as two qualitativelydifferent forms of social stratification. Two cases serve to illustrate thelimitations and dangers that result from neglecting the conceptual andempirical work undertaken by sociologists: the study of intergenerationalsocial mobility by economists and the study of the consequences of so-cial inequality for health and related social problems by epidemiologists.
Palabras claveDesigualdad social Clase social Esta-tus social Movilidadsocial Economistas Epidemilogos
ResumenLas cuestiones relativas a la desigualdad social han adquirido recientemen-te una nueva centralidad poltica en muchas sociedades occidentales. Sinembargo, las aproximaciones a la desigualdad social que han realizado lossocilogos han sido ignoradas, especialmente en el trabajo de economis-tas y epidemilogos. Los principales rasgos del enfoque sociolgico sonel nfasis en la desigualdad entendida en un sentido relacional ms quesimplemente atributivo y la distincin entre la clase social y el estatus socialcomo dos formas cualitativamente diferentes de estratificacin. Se presen-tan dos casos para ilustrar las limitaciones y los peligros que resultan deignorar el trabajo emprico y conceptual de los socilogos: el estudio de loseconomistas sobre la movilidad social intergeneracional y el estudio de los
epidemilogos acerca de las consecuencias de la desigualdad social parala salud y otros problemas sociales relacionados.
7/29/2019 2012 Goldthorpe Sociologia de La Desigualdad
2/15
Reis 137, enero-marzo 2012, pp. 201-216
202 Back to Class and Status: Or Why a Sociological View of Social Inequality Should Be Reasserted
one generation to the next had greatly weake-
ned, as indicated by the quite low correlation
existing between parents incomes and theincomes of their children (e.g. Becker and To-
mes, 1979). Crucial factors in these develop-
ments, it was argued, were the increasingimportance in economic growth of human
capital relative to physical capital and the ex-
pansion and reform of educational systems.
From the side of sociology, analogous claims
were put forward by theorists of industrialism
and post-industrialism. It was envisaged that
older, more rigid, forms of social stratificationwere giving way to a rather amorphous so-cio-economic hierarchy, within which indivi-
duals positions were determined far less byascription than by their own educationaland occupational achievement (e.g. Par-sons, 1967, 1971; Treiman, 1970).
From the 1980s onwards, such meliorist
views became more contested among eco-
nomists and sociologists alike as evidence
emerged that social inequalities showed in
many respects a strong persistence if not a
tendency to increase. And, as evidence of
this kind has continued to accumulate, issues
of inequality have in more recent times assu-
med a new political centrality in many wes-
tern societies. However, while sociologists
have for the most part still focussed their at-
tention on relatively long-term trends in in-
equality, economists interests have shown amuch sharper shift towards the present-day
- with the result that their concerns have of-
ten come to have a greater resonance with
those prevailing outside of academia.
The new interest in inequality among eco-
nomists has in fact been prompted in several
different ways. First of all, the most widely
apparent increase in inequality in the recent
past has been in the distribution of incomes.
While some significant cross-national variationexists in this respect, it is at all events clear
enough that the Kuznets curve no longer holds
(Atkinson, 2008). Further, technical improve-
ments in the study of intergenerational income
mobility have led to results showing that this
is a good deal more restricted than was pre-
viously believed. Correlations between pa-
rents and childrens incomes, which wereonce thought to be as low as 0.2, are now
generally estimated at 0.4 or even higher
(Bowles, Gintis and Osborne Groves, eds.,2005; Bjrklund and Jntti, 2009). At the sametime, evidence of increasing income inequality
and a more realistic view of income mobility
have encouraged new thinking about the con-
sequences of economic inequality. It was once
widely supposed that a trade-off had to bemade between greater equality, on the one
hand, and economic efficiency and growth, onthe other (e.g. Okun, 1975). But economists
have now become more prepared to considerthe possibility that inequality may itself have
adverse effects on efficiency and growth (seee.g. Aghion, Caroli and Garca-Pealosa,1999) as well as on various aspects of indivi-
dual welfare and, in particular, on health.
In this latter regard, an interesting enga-
gement has of late developed between eco-
nomists and epidemiologists. There is gene-
ral agreement that economically more
advantaged people have on average better
health than those less advantaged. But disa-
greement tends to arise over whether, over
and above such effects at the individual level,
there are also population effects of inequality
- or, one could say, contextualeffects: that is,
effects that mean that in more equal societies
individuals at all economic levels have better
health than do their counterparts in less equal
societies. Some epidemiologists, such as Mi-
chael Marmot (2004) and Richard Wilkinson
(1996, 2005; Wilkinson and Pickett, 2009),
would claim that for the economically advan-
ced societies of today such contextual effects
are in fact well-established. Their position is
exemplified by graphs such as that of Figure1 which relates income inequality to life ex-
pectancy in 23 advanced societies. As can
be seen, there is a tendency across these so-
cieties for life expectancy to fall as income
inequality increases. If such analyses are ac-
cepted, one implication would then be that in
7/29/2019 2012 Goldthorpe Sociologia de La Desigualdad
3/15
Reis 137, enero-marzo 2012, pp. 201-216
John H. Goldthorpe 203
highly unequal societies, such as the US or
Portugal, the health of well-off people may be
no better than, or even worse than, the health
of poorer people in more equal societies,
such as Japan or Sweden.
However, most economists would be
sceptical of the Marmot-Wilkinson position.
They would argue that correlations such as
those reflected in Figure 1 are not statistically
robust. They may depend on the particular se-lection of countries made; they can disappear
when further variables, such as education or
ethnicity, are introduced into the analysis; and,
most seriously perhaps, they often fail to show
up if a dynamic, over-time perspective is taken
(see e.g. Mellor and Milyo, 2001; Deaton,
2002; Leigh, Jencks and Smeeding, 2009).1
1 It should be added that a number of other epidemio-logists would also take issue with Marmot and Wilkinson
on whether a population effect of social inequality on
Now, this engagement is, in itself, very
much to be welcomed. It is good that in eco-
nomics the study of inequality has, in the
words of Tony Atkinson (1997), been broughtin from the cold; and it is good that the fo-rays of epidemiologists into the social scien-
ces should be viewed seriously, albeit critica-
lly. But what, for sociologists, must be
disturbing in all of this is the way in which
their distinctive - and, I wish to maintain,
highly relevant - contributions to the study of
social inequality are almost entirely disregar-
ded.
This is the result in part, as I have sugges-
ted, of sociologists not participating as acti-
vely as they might have done in debates
about inequality today as opposed to deba-
health is invariably present in advanced societies. See
e.g. Lynch et al. (2000, 2004).
FIGURE 1. Income Inequelity and Life Expectancy
Source: Wilkinson and Pickett (2009).
7/29/2019 2012 Goldthorpe Sociologia de La Desigualdad
4/15
Reis 137, enero-marzo 2012, pp. 201-216
204 Back to Class and Status: Or Why a Sociological View of Social Inequality Should Be Reasserted
tes about long-run tendencies. But it is also
the result, I believe, of both economists and
epidemiologists remaining remarkably fixedwithin their disciplinary paradigms, and thus
being either unaware of, or uncomfortable
with, sociological concepts, and with relatedempirical enquiry that has a very direct bea-
ring on current debates.
In consequence, I would argue, a good
deal of recent research on inequality by
economists and epidemiologists has been,
at best, unduly limited and, at worst,
flawed. In what follows, I develop this argu-ment by drawing on a range of recent con-
ceptual, theoretical and empirical contribu-
tions by sociologists with two aims in mind:
first, to bring out what is chiefly distinctiveabout the sociological approach to inequa-
lity; and, second, to show, on the basis of
two particular examples, how the disregard
of this approach, by economists and by
epidemiologists respectively, has led to se-
rious difficulties.
The sociological approach to socialinequality
In discussing inequality, economists concen-
trate their attention on income and wealth,
and may also refer to inequalities in educatio-
nal attainment which, via human capital
theory, they treat as a major determinant of
income. In these respects, they are concer-
ned with inequality in what might be called an
attributionalsense. Income, wealth, educa-
tion are attributes of individuals, of whichthey have more or less. In contrast, sociolo-
gists tend to discuss inequality in terms of
social class or social status, and thus to treat
inequality in arelationalsense: i.e. in terms of
social relationships within which individuals
are more or less advantaged or disadvanta-
ged.
Thus, in what is, I believe, becoming an
increasingly common view, class is taken to
be defined by social relations within labourmarkets and productive units (Erikson and
Goldthorpe, 1992, ch. 2; Goldthorpe, 2007,
vol. 2, ch. 5; McGovern et al., 2007, ch. 3).
Initial distinctions therefore arise between
employers, self-employed workers and em-
ployees; and then, among employees, fur-
ther distinctions are made according to theform of their relations with employers, as
embodied in the terms of their employment
contracts (implicit as well as explicit). For
example, a crucial distinction here is that
between salaried employees, on the one
hand, and, on the other hand, wage-workers
who are employed on some kind of piece- or
time-rate basis. Such an approach to class
can now be made operational for purposes
of empirical research through classificationswell-established in the sociological literature
- and in some cases adopted by national
statistical offices, and including now a pro-totype European Union Socio-economic
Classification (Rose and Harrison, eds.,
2009).
Economists may say that they are prima-
rily interested in economic, rather than wider
social inequality. But there are good grounds
for maintaining that the concept of class, un-
derstood in the way indicated, does in fact
lead to a more comprehensive view of eco-
nomic inequality than does a focus on inco-
me, and especially on current income, alone.
It can be shown that individuals in different
class positions differ systematically not only
in their levels of income but in at least three
further ways (Goldthorpe and McKnight,
2006; Chan and Goldthorpe, 2007): first, intheir degree of income security; second, in
theirshort-term income stability; and, third, in
their longer-term income prospects. Thus,
members of the professional and managerial
salariat, as well as having generally higher
incomes than wage-workers, have lower
risks of loss of income as a result of unem-
ployment, have incomes less dependent on
variable rates of pay, and, perhaps most im-
portantly, have incomes that continue to rise
for far longer over the course of their working
lives on account of incremental salary scales
7/29/2019 2012 Goldthorpe Sociologia de La Desigualdad
5/15
Reis 137, enero-marzo 2012, pp. 201-216
John H. Goldthorpe 205
and relatively well-defined promotion and ca-reer opportunities.2
An emphasis on the relational, rather than
simply the attributional, aspects of inequality
is, then, one way in which the sociological
approach is distinctive. A second way lies in
recognition of the fact that the structuring of
inequality - or social stratification - is morethan one-dimensional. At least since the time
of Max Weber (1922/1968), sociologists have
thought about inequality not only in terms of
class but also in terms of another relational
concept, that of status.
The status order - or hierarchy - is one
formed by social relations of superiority,equality and inferiority that reflect prevailingevaluations of social honour or worth. In ear-
lier societies, status typically attached to as-
cribed characteristics - in particular, to birthor descent. In present-day societies statusmore often attaches to social positions - in
particular, occupations - although still also to
ascribed characteristics such as race and
ethnicity (Laumann, 1966). The most imme-
diate way in which the status order is expres-sed is in patterns of intimate association,
such as close friendship and marriage (or co-
habitation). Status equals are those who eat
together and sleep together. But differences
in status are also expressed in lifestyles of
differing distinction that are seen as appro-priate to different status levels.
Just as there are now good instruments
available for the measurement of class, so
too there are good instruments available forthe measurement of status as understood as
above, and especially as based on occupa-
tion (see e.g. Chan, 2010).3 And what can
2 Members of classes intermediate between the profes-
sional and managerial salariat and the working class can
be shown also to be in various ways intermediate as
regards income security, stability and prospects (Gol-
dthorpe and McKnight, 2006; McGovern et al., 2007: ch.
3 esp.)3 Note that these scales based on the occupational
structuring of patterns of intimate association, such as
then be shown is that while, as one might
expect, the positions that individuals hold
within class structures and status hierar-
chies are correlated, the correlation tends to
be only moderate. Class and status incon-
sistencies clearly arise. In other words,class and status have to be regarded as two
qualitatively different forms of social stratifi-cation.
It is then in this connection that a socio-
logical approach to inequality differs most
sharply from that typically found in epide-
miology. Epidemiologists have been greatly
concerned with social inequalities - or
what they call social gradients - in morta-lity, morbidity and other aspects of health.
But while they show great care and sophis-
tication in the measurement of these de-
pendent variables of their analyses, epide-
miologists measurement of socialinequality is for the most part remarkably
casual and ad hoc. The underlying as-
sumption seems to be that inequality is
essentially one-dimensional: there is a sin-
gle social hierarchy and individuals posi-tions in this hierarchy can be determinedby a range of indicators that are more or
lessinterchangeable: for example, income,
education or occupational level as deter-
mined by various different criteria.
However, this assumption of unidimensio-
nality is unwarranted. Although, as noted at
the outset, some sociologists have drifted
away from a Weberian position to envisage
social stratification in terms of a single socio-economic hierarchy, a wide-ranging literatu-re does in fact exist to show the continuing
importance of distinguishing between class
and status. To provide just one illustration,
Table I reveals how, in Britain, class and sta-
tus relate to socio-political attitudes as mea-
close friendship or marriage, are to be distinguished both
from scales of occupational prestige and from scales of
socio-economic status, based typically on data on oc-cupational earnings and education. See further Bukodi,
Dex and Goldthorpe (2011).
7/29/2019 2012 Goldthorpe Sociologia de La Desigualdad
6/15
Reis 137, enero-marzo 2012, pp. 201-216
206 Back to Class and Status: Or Why a Sociological View of Social Inequality Should Be Reasserted
sured by two well-established scales: one
capturing individuals positions on a Right-Left dimension, the other their positions on
an Authoritarian-Libertarian dimension
(Evans, Heath and Lalljee, 1996).
As regards the Right-Left scale, it can be
seen that class, and also income, are the im-
portant differentiating factors and on a pat-
tern that would be expected - i.e. working-
class and low income individuals are the
most left-wing - while the effect of status is
not significant.4 In contrast, with the Authori-
tarian-Libertarian scale, it is status, and also
education, that count - authoritarianism de-
clines with status and higher education - whi-
le class and income are of little consequence.
4 Education also has significant effects but on a curvili-near pattern. As can be seen, individuals with the highest
and lowest levels of educational qualification are mostleft-wing, while those with intermediate levels are most
right-wing.
TABLE 1. Determinants of political attitudes on Right-Left and Authoritarian-Libertarian scales, OLS regression,
British Social Attitudes Survey, 2002 (boldface coefficients are significant at 1% level, two-tailed
tests)
Right-Left Authoritarian-Libertarian
Age 0.005 0.030
Gender (male ref.)
Female 0.605 0.136
Income (44k 2.153 -0.111
Qualifications (none ref.)
CSE 0.472 0.193
O-levels 1.039 0.261A-levels 1.090 0.726
Sub-degree 1.089 0.822
Degree 0.153 -3.223
Class (I, Prof. and manag., higher, ref.)
II Prof. and manag., lower 0.873 0.020
III Routine nonmanual 1.233 0.004
IV Small proprietors 0.021 0.138
V Technicians and supervisors 1.553 0.082VI Skilled manual 1.551 0.340
VII Nonskilled manual 1.732 0.130
Status 0.684 1.381
Constant 11.711 21.363
N 2.421 2.441
R2 0.130 0.170
Source: Adapted from Chan and Goldthorpe (2007: Table 7)..
7/29/2019 2012 Goldthorpe Sociologia de La Desigualdad
7/15
Reis 137, enero-marzo 2012, pp. 201-216
John H. Goldthorpe 207
Taking a one-dimensional view of social stra-
tification would then simply obscure theserather important findings.
Having set out what I take to be the essen-
tials of the sociological approach to social in-
equality, I now turn to cases which, I believe,
illustrate the dangers that arise as a result of
economists and epidemiologists largely ne-
glecting the conceptual basis of this approach
and the research findings that have followedfrom it. In regard to the work of economists, the
case I wish to take is that of social mobility.
Economists and social mobility
In recent years, social mobility has in severalcountries become a central political issue. It
is, moreover, one that is now attracting the
attention of international organisations, in
particular the OECD. However, what is re-
markable about the reports that the OECD
has produced (e.g. Causa, Dantan and Jo-
hansson, 2009; OECD, 2010: ch. 5) is that
they are written entirely by economists, who
never move beyond the limits of their own
disciplinary paradigm and make little or noreference to the very substantial body of
work previously done in this field by sociolo-gists. Social mobility is simply equated with
income or earnings mobility, and without any
attempt at explanation or justification. I willaim to show why this is unfortunate by con-
sidering aspects of a debate which I and se-
veral others have been carrying on with eco-
nomists concerning social mobility in Britain.
The economists involved, chiefly Jo Blan-den and her colleagues, investigated income
mobility using data from two birth-cohort stu-
dies covering all children born in Britain in
one week in 1958 and in one week in 1970.
They related the earnings of these children,
when they were in their early 30s, to the inco-
mes of their families when they were in their
adolescence. They then claim to show a
much stronger association between family
income and childrens later earnings for the1970 cohort than for the 1958 cohort: i.e. the-
re is less income mobility in the later cohort
than in the earlier one (Blanden et al., 2004).
This research has had an enormous impact in
media and political circles. It has led to a wi-
despread belief that social mobility in Britain
is in sharp decline. All three leading politicalparties have published reports dealing with
the - supposed - problem, and the present
coalition government has followed the pre-
vious New Labour government in making in-
creased social mobility one of its major ob-
jectives.
However, Michelle Jackson, and I (Gol-
dthorpe and Jackson, 2007) have used the
same birth cohort data as the economists in
order to investigate intergenerational class
mobility and, at the same time, to distinguish
betweenabsolute andrelative rates of mobi-
lity: that is, between mobility rates as measu-
red in simple percentage terms and mobility
rates as measured by odds ratios showing
the degree of association existing between
parental class and childs class when consi-derednetof all class structural change (on
this distinction, see further Erikson and Gol-dthorpe, 1992: 54-9; Breen, 2004). So far as
relative mobility rates - or what may be called
social fluidity - are concerned, we find, incontrast to the economists, no significantchange between the 1958 and 1970 cohorts.5
Moreover, together with Colin Mills, I have
5 Until recently, economists appear to have treated ques-
tions of mobility in innocence of the distinction between
absolute and relative rates. Where income is analysed in
terms of movement between income quantiles, the analy-
sis is of course relativised from the start. But wheremobility is assessed in terms of elasticities - i.e. throughthe regression of childrens income on parents income- effects of the net association existing between these
variables and of differences in their distributions will tend
to be confounded. Bjrklund and Jntti, two economistsworking in this field with an unusually good knowledgeof the sociological literature, have recently proposed
(2009) using correlations rather than elasticities where
the focus of interest is on change in mobility over time
or on variation across societies. In fact, Blanden et al.
(2004) use all of the above methods and in each casealike find a decline in mobility across the two cohortsthey study.
7/29/2019 2012 Goldthorpe Sociologia de La Desigualdad
8/15
Reis 137, enero-marzo 2012, pp. 201-216
208 Back to Class and Status: Or Why a Sociological View of Social Inequality Should Be Reasserted
analysed data from 13 national sample sur-
veys, carried out between 1972 and 2005,
which give a far better basis than do the co-
hort studies for assessing population trends
in mobility (Goldthorpe and Mills, 2008). On
this basis, the Goldthorpe-Jackson finding issupported in that only very slight, and quite
trendless, fluctuations in relative rates ofclass mobility show up over the entire period
covered.
Now there is, of course, no reason why
studies of income mobility and of class mo-
bility should show the same results. But it is
at the same time of interest to look further
into how it is that they differ. So in yet anotherpaper, Robert Erikson and I (Erikson and Gol-
dthorpe, 2010) have returned to the birth-
cohort data and have carried out further
analyses using only those members of the
two cohorts who could be included in both
the economists analysis and in the Goldthor-pe-Jackson analysis: i.e. those individuals for
whom we have data on both income and
class mobility.
In Table II I show some results from a lo-
glinear modelling exercise relating to mobility
tables for these individuals based on income
quintiles and on five social classes. Note, firstof all, that all previous findings are fully con-firmed. For class mobility, the constant social
fluidity (CSF) model gives an acceptable fitfor both men and women, on which the UNI-
TABLE 2. Results offitting independence, constant socialfluidity (CSF) and uniform difference (UNIDIFF)
modelsa to income quintile andfive class mobility tables for 1958 and 1970 British birth cohorts
Table Modela G2 Pb DI
men (N = 3415)
Indep. 248.7 0.00 10.2
Income quintile CSF 19.6 0.24 3.1
UNIDIFF 9.5 0.85 1.9 (1970) = 1.54
Indep. 410.0 0.00 13.2
Five classc CSF 12.7 0.70 2.3
UNIDIFF 12.5 0.64 2.3 (1970) n.s.
women (N= 3009)
Indep. 192.0 0.00 9.4
Income quintile CSF 31.8 0.01 4.1UNIDIFF 21.0 0.13 3.5 (1970) = 1.70
Indep. 291.8 0.00 12.2
Five classc CSF 23.9 0.09 2.8
UNIDIFF 22.6 0.09 2.6 (1970) n.s.
a Independence: log Fijk =+ iO
+ jD
+ kC
+ ikOC
+ jkDC
CSF: log Fijk =+ iO
+ jD
+ kC
+ ikOC
+ jkDC
+ ijOD
UNIDIFF: log Fijk =+ iO
+ jD
+ kC
+ ikOC
+ jkDC
+ kXij
(O = origin, D = destination, C = cohort)
b Degrees of freedom are: Independence model, 32; CSF model 16, UNIDIFF model, 15.
c
1: Higher professionals and managers; 2: Lower professionals and managers; 3: Higher routine nonmanual and technicians;4: Skilled manual; 5: Lower routine nonmanual and nonskilled manual.
Source: Adapted from Erikson and Goldthorpe (2010: Tables II and III).
7/29/2019 2012 Goldthorpe Sociologia de La Desigualdad
9/15
Reis 137, enero-marzo 2012, pp. 201-216
John H. Goldthorpe 209
DIFF model, proposing some uniform changein the level of the log-odds ratios defining re-lative mobility, does not improve. But for in-
come mobility the UNIDIFF model does im-
prove significantly on the CSF model for bothmen and women; and the parameters retur-
ned for the 1970 cohort are in both cases
positive, indicating lower fluidity for this co-hort than for the 1958 cohort, just as the eco-
nomists found. Specifically, the model im-
plies that all log-odds ratios underlying theincome mobility tables increase by a factor of
1.54 for men and 1.70 for women.
However, one should further note the G2
statistics returned under the independence
model. For men and women alike, these are
much higher for class mobility than for inco-
me mobility - suggesting, that is, that a stron-
ger intergenerational association prevails in
the case of class than in the case of income.
To test this possibility more directly, Erikson
and I calculate the global log-odds ratios that
result from partitioning the 5 x 5 income and
class mobility tables into successive 2 x 2
tables, and we then average these ratios. The
results are shown in Table III.
It can be seen that in all cases, with both
the unweighted and weighted6 averages, the
6 The weights used are the inverted variances of the
log-odds ratios.
odds ratios are higher - i.e. there is a strongerintergenerational association - in the case of
class than in the case of income. This differen-
ce is very marked - and highly statistically sig-
nificant - in the 1958 cohort, while diminishingin the 1970 cohort, especially in the case of
men (see further Cox, Jackson and Lu, 2009).
In our paper, Erikson and I argue that the-
re are in fact major problems with the income
data that the economists use, which in them-
selves make the finding of declining mobilitybetween the two cohorts a questionable one.
However, for present purposes this issue can
be left aside. The main point to be made here
is that even if there were such a decline in
income mobility, it was one that took place
within a class mobility regime that shows
both less fluidity overallandgreater stabilityover time. In other words, it is the class mo-
bility regime that would appear more fully and
reliably to capture the continuity in economicadvantage and disadvantage that persists
across generations - as might indeed be ex-
pected given the relationship between class
and income that was previously described.
Studies that are based on income mobility
alone - such as those of the OECD - would
seem therefore to run a serious risk of under-
estimating the propensities for economic
immobility.
In this connection, it is relevant to add
that Erikson and I further investigate the rela-
TABLE 3. Averages of log odds ratios obtained from 2 x 2 partitioning of quintile income andfive class mobility
tables
Men Women
Averages Income Clases Income Class
1958 0.60 1.17 0.48 0.84
Unwieghted
1970 0.97 1.03 0.79 1.00
1958 0.58 1.12 0.46 0.84
Weighted
1970 0.94 1.02 0.80 0.99
Source:Adapted from Erikson and Goldthorpe (2010: Table IV).
7/29/2019 2012 Goldthorpe Sociologia de La Desigualdad
10/15
Reis 137, enero-marzo 2012, pp. 201-216
210 Back to Class and Status: Or Why a Sociological View of Social Inequality Should Be Reasserted
tionship in the two British birth cohorts bet-
ween childrens educational attainment andboth family income and parental social class.
The results we obtain are on much the same
lines as those shown for mobility in Tables 2
and 3. The association between childrenseducational attainment and family income
does strengthen across the two cohorts - as
the economists have in fact reported (Blan-
den and Machin, 2004; Blanden, Gregg and
Machin, 2005) - while the association bet-
ween childrens educational attainment andparental class shows no significant change.But, in both cohorts alike, the association of
educational attainment with class is stronger
than the association with family income (Erik-son and Goldthorpe, 2010: Tables VII and
VIII). Again, therefore, a focus simply on inco-
me rather than on class would seem likely to
lead to an underestimation of the impact of
social background on childrens life-chances.Relational inequality is more consequential
than attributional inequality.
I move on now to the case I want to con-
sider in regard to epidemiologists, which
concerns the consequences of social inequa-
lity.
Epidemiologists and the consequences
of social inequality
I earlier showed, in Figure 1, a graph taken
from the work of Wilkinson and Pickett (2009)
relating income inequality to life expectancy
in a number of advanced societies. Figure 2
shows another graph taken from the samework that similarly relates income inequality
to a more general index of health and various
social problems.
It is evident that the cross-societal pattern
in Figure 2 is essentially the same as that in
Figure 1. As income inequality increases,
health and social problems get worse.
However, what is important to note is that
Wilkinson and Pickett do not believe that it is
the direct - that is, the material - consequen-
ces of income inequality that are here revea-
led. Rather, they make the assumption I ear-
lier referred to that the structure of social
inequality is essentially one-dimensional, and
they thus take income inequality as being a
good indicator of what for themis the crucial
causal factor underlying their graph, namely,status inequality. It is status inequality - and
the insecurity and anxiety, the damaged self-
esteem and lack of trust that it generates -
that impact on health and on individual and
social well-being more generally. The psy-
chological stress created by status inequality
exerts its negative effects in two ways: most
immediately, on health, through various neu-
roendocrine mechanisms but also, more per-
vasively, through inducing harmful behaviourin the form of smoking, over-eating, sexual
promiscuity or violence.
This view of the effects of status inequali-
ty is in itself controversial.7 But the problem I
wish to bring out here is one that lies further
back in Wilkinson and Picketts argument:that is, with the assumption that social strati-
fication is one-dimensional and that the de-gree of status inequality in a society can the-
refore be reliablyinferred from the degree ofincome inequality. As I have already noted,
the available evidence indicates that the sta-
tus and class positions of individuals are only
moderately correlated, and the same is true
of their status positions and their income le-
vels (Chan, 2010). It would therefore be dan-
gerous to suppose that any different situation
exists at the societal level, and there is in fact
one case in particular that underlines this po-
int - that of Japan.
7 The same epidemiologists referred to in n.1 above who
doubt if a population effect of inequality on health is
always present would also believe (Lynch et al., 2000,
2004) that, where such an effect is present, it operates
less through the psychological effects of status inequa-
lity than through more direct material effects, although
not those of income inequality alone. Societies that have
low income inequality, it is argued, also tend to have
more health-supportive infrastructures in the form of
health services, occupational health and safety regula-tion, environmental controls, unemployment benefits andback-to-work programmes etc.
7/29/2019 2012 Goldthorpe Sociologia de La Desigualdad
11/15
Reis 137, enero-marzo 2012, pp. 201-216
John H. Goldthorpe 211
As can be seen from Figure 2, Japan is
at the extreme good end of Wilkinson andPicketts graph. Japan has comparativelylow income inequality - largely in fact on ac-
count of low earnings inequality - and Japan
also scores very well on the health and so-
cial problems index. However, among stu-
dents of comparative social stratification,Japan is noted for having a very strongly de-
fined status hierarchy, and one that is to anunusual degree formalised and embodied in
language through the extensive use of hono-
rifics. To quote a leading expert, Harold Ker-bo (2003: 479-80): the Japanese seem ob-sessed with ranking and hierarchy; ineveryday life it is only once status relevantmarkers have been established thatthe business of eating, talking, drinking, or
whatever can proceed in an orderly manner
that is unlikely to offend someone who ex-
pects greater status deference. And Kerboin fact goes on to suggest (2003: 509-12)
that for individuals in high-level positions,
as, say, in the corporate world, high status
is itself a reward that helps to compensate
for their - relatively - low material rewards.
The Japanese case does, therefore, create
serious difficulties for Wilkinson and Pickett- of which they seem largely unaware - and,
more generally, underlines the need for epi-
demiologists to move towards a more so-
phisticated, sociological way of thinking
about social inequality.8
8 There are in fact some welcome indications that a movein this direction is beginning. See e.g. Geyer et al. (2006)
and especially Galobardes et al. (2006).
FIGURE 2. Income Inequality and Health and Social Problems
Note: Health and social problems are closely related to inequality among rich countries.
7/29/2019 2012 Goldthorpe Sociologia de La Desigualdad
12/15
Reis 137, enero-marzo 2012, pp. 201-216
212 Back to Class and Status: Or Why a Sociological View of Social Inequality Should Be Reasserted
Finally, to make this same point more
positively, I would note recent research by
two sociologists, Jenny Torssander and Ro-
bert Erikson (2010), who have trespassed
on epidemiologists territory largely in reac-tion against their one-dimensional view of
stratification. Torssander and Erikson useregistration data, covering the entire Swe-
dish population, to investigate inequalities
in the risk of mortality by class and status,
as relational variables, and also by incomeand education, as attributional variables.
Table IV shows some of the results they ob-
tain.
When the four explanatory variables are
taken separately, as in the left-hand panel of
the table, each reveals a gradient in the risksof death of the kind one might expect, with
the one exception of income in the case of
women. So, one might think, these measures
of social inequality are, more or less, inter-
changeable. However, when - with no pro-
blems of collinearity - all four variables are
brought into the analysis together, as in the
TABLE 4. Results of bivariate and multivariate Cox regressions of relative risk of death on class, status, income
and education, Swedish men and women aged 35-59 in 1990 (boldface coefficients are significant at
5% level)
Bivariate Multivariate
men women men women
Classa
1 1 1 1 1
2 1.17 0.98 1.01 0.94
3 1.37 1.18 1.07 1.01
4 1.61 1.18 1.09 0.90
5 1.87 1.36 1.18 1.03
Statusb
Quintile 1 1 1 1 1
Quintile 2 1.19 1.05 1.04 1.02
Quintile 3 1.36 1.23 1.03 1.11
Quintile 4 1.69 1.27 1.04 1.09Quintile 5 1.80 1.49 1.09 1.28
Income (from work, average 1981-1989)
Quintile 1 1 1 1 1
Quintile 2 1.18 1.14 1.07 0.99
Quintile 3 1.38 1.12 1.14 0.95
Quintile 4 1.55 1.07 1.23 0.89
Quintile 5 2.29 1.14 1.81 0.92
Education
Higher tertiary 1 1 1 1
Lower tertiary 1.14 1.04 1.06 1.06
Higher secondary 1.26 1.23 1.13 1.17Lower secondary 1.67 1.34 1.28 1.24
Compulsory only 1.76 1.48 1.27 1.30
a 1: Higher professionals and managers; 2: Lower professionals and managers; 3: Higher routine nonmanual and technicians;
4: Skilled manual; 5: Lower routine nonmanual and nonskilled manual.
b Scale based on occupational homogamy/heterogamy.
Source: Adapted from Torssander and Erikson (2010: Table 4).
7/29/2019 2012 Goldthorpe Sociologia de La Desigualdad
13/15
Reis 137, enero-marzo 2012, pp. 201-216
John H. Goldthorpe 213
right-hand panel of the table, a different pic-
ture emerges. Education remains a major fac-
tor in risks of death for men and women alike.
But while status has still some importance for
women, it has little for men; and while class
and income still matter for men, neither nowmatters for women. In other words, any one-
dimensional approach to social stratificationwould, on this evidence, quite fail to bring out
the complexity that apparently exists in the
social generation of the risks of death.
This claim is indeed nicely supported by
further results that Torssander and Erikson
(2009) have obtained when they consider
spouse (or partner) effects on risk of death.
These results could be summed up as fo-
llows. For a woman who wants a long life - at
least in Sweden - the best thing for her to do
is to get together with a high class man with
a large income; but for a man who wants a
long life, the best thing for him to do is to get
together with a high status woman with a
good education.9
CONCLUSIONS
In this paper, I have aimed to reassert the im-
portance of a sociological approach to the
understanding of social inequality. At a con-
ceptual level, I have noted, first, the emphasisthat sociologists, in contrast to most econo-
mists, would give to the relational as distinct
from the simply attributional aspects of in-
equality; and, second, sociologists aware-
ness, in contrast to most epidemiologists, ofthe multidimensional nature of the structuring
of social inequality and in particular of class
and status as two qualitatively different forms
9 Torssander and Erikson (2009) also broach the question
of possible relationships between different aspects and
forms of social inequality and the risks of death from
different causes. Further investigation of this question
might help in resolving current disagreements among
epidemiologist (see n. 7 above) about the relative impor-tance of the impact on health of the psychological andmaterial implications of inequality.
of social stratification. I have then furthersought to show how, in two specific areas,sociological research has revealed difficultiesin work by economists and epidemiologists,
respectively, in which these conceptual in-
sights have been disregarded.
I believe that in the years ahead - in con-
sequence of the present economic crisis and
indeed of the form of neo-liberal political eco-
nomy from which the crisis stems - the gene-
ral tendency will be for social inequality, in all
its forms, to widen yet further. In turn, public
and political concern with the issues that ari-
se can be expected to become more pres-
sing. Recent contributions of economists and
epidemiologists have been of great value in
helping to refocus social scientific researchon these issues of inequality. But it is now
urgent that sociologists should engage more
fully and more forcefully than hitherto in rele-
vant empirical inquiry and analysis, and seek
explicitly to demonstrate the distinctive ad-
vantages that their approach can provide.
REFERENCES
Aghion, Philippe Eve Caroli and Cecilia Garca-Pea-losa (1999): Inequality and Economic Growth:the Perspective of the New Growth Theories,Journal of Economic Literature, 37: 1615-1660.
Atkinson, Anthony B. (1997): Bringing Income Dis-tribution in from the Cold, Economic Journal,107: 297-321.
(2008): The Changing Distribution of Earnings in
OECD Countries, Oxford: Oxford UniversityPress.
Becker, Gary and Nigel Tomes (1979):An EquilibriumTheory of the Distribution of Income and Interge-
nerational Mobility,Journal of Political Economy,87: 1153-89.
Bjrklund, Anders and Markus Jntti (2009): Inter-generational Income Mobility and the Role of
Family Background, in W. Salverda, B. Nolan andT. Smeeding (eds.), Oxford Handbook of Econo-
mic Inequality, Oxford: Oxford University Press.
Blanden, Jo and Stephen Machin (2004): Educatio-nal Inequality and the Expansion of Higher Edu-
7/29/2019 2012 Goldthorpe Sociologia de La Desigualdad
14/15
Reis 137, enero-marzo 2012, pp. 201-216
214 Back to Class and Status: Or Why a Sociological View of Social Inequality Should Be Reasserted
cation, Scottish Journal of Political Economy, 51:230-49.
, Alissa Goodman, Paul Gregg and Stephen Ma-chin (2004): Changes in Intergenerational Mobi-lity in Britain, in M. Corak (ed.), Generational
Income Mobility in North America and Europe,Cambridge: Cambridge University Press.
, Paul Gregg, and Stephen Machin (2005): Edu-cational Inequality and Intergenerational Mobilityin S. Machin, and A. Vignoles (eds.), Whats the
Good of Education?, Princeton: Princeton Uni-
versity Press.
Bowles, Samuel, Herbert Gintis and Melissa Osborne
Groves (eds.) (2005): Unequal Chances: Family
Background and Economic Success, New York:
Russell Sage Foundation and Princeton: Prince-
ton University Press.
Breen, Richard (2004). Statistical Methods of Mobi-lity Research in R. Breen (ed.), Social Mobility inEurope, Oxford: Oxford University Press.
Bukodi, Erzsbet, Shirley Dex and John H. Goldthor-pe (2011): The Conceptualisation and Measure-ment of Occupational Hierarchies: A Review, a
Proposal and some Illustrative Analyses, Qualityand Quantity, 45: 623-639.
Causa, Orsetta, Sophie Dantan andsa Johansson(2009): Intergenerational Social Mobility in Euro-pean OECD Countries, OECD Economics De-
partment Working Papers, N 709.
Chan, Tak-Wing (2010): The Social Status Scale: ItsConstruction and Properties, in T.-W. Chan (ed.),Social Status and Cultural Consumption, Cam-
bridge: Cambridge University Press.
and John H. Goldthorpe (2007): Class and Sta-tus: The Conceptual Distinction and its Empirical
Relevance, American Sociological Review, 72:
512-532.
Cox, D. R., Michelle Jackson, M. and Shiwei Lu
(2009): On Square Ordinal Contingency Tables:A Comparison of Social Class and Income Mo-
bility for the Same Individuals, Journal of theRoyal Statistical Society(Series A), 172: 483-493.
Deaton, Angus (2002): The Convoluted Story of In-ternational Studies of Inequality and Health,International Journal of Epidemiology, 31: 546-
549.
Erikson, Robert and John H. Goldthorpe (1992): TheConstant Flux: A Study of Class Mobility in Indus-
trial Societies, Oxford: Clarendon Press.
and (2010): Has Social Mobility in Britain De-creased? Reconciling Divergent Findings on In-
come and Class Mobility, British Journal of So-ciology, 61: 211-230.
Evans, Geoffrey, A. Heath, and Lalljee, M. 1996:
Measuring Left-Right and Libertarian-Authorita-rian Values in the British Electorate, British Jour-nal of Sociology, 47: 93-112.
Galobardes, B., Shaw, M., Lawlor, D., Lynch, J. and
Davey Smith, G. (2006): Indicators of Socioeco-nomic Position (Parts 1 and 2), Journal of Epi-demiology and Community Health,60: 7-12 and
95-101.
Geyer, S. Hemstrm, , Peter, R. and Vger, D.(2006): Education, Income, and OccupationalClass Cannot be Used Interchangeably in Social
Epidemiology. Empirical Evidence against a Com-mon Practice, Journal of Epidemiology andCommunity Health, 60, 804-810.
Goldthorpe, John H. (2007): On Sociology, 2 ed., 2
vols. Stanford: Stanford University Press.
and Abigail McKnight (2006): The Economic Ba-sis of Social Class, in S. L. Morgan, D. B. Grus-ky and G. S. Fields (eds.), Mobility and Inequality:
Frontiers of Research in Sociology and Econo-
mics, Stanford: Stanford University Press.
and Michelle Jackson (2007): IntergenerationalClass Mobility in Contemporary Britain: Political
Concerns and Empirical Findings, British Journalof Sociology, 58: 526-546.
and Colin Mills (2008): Trends in Intergeneratio-nal Class Mobility in Modern Britain: Evidence
from National Surveys, 1972-2005, National Ins-titute Economic Review, 205: 1-18.
Kerbo, Harold R. (2003): Social Stratification and Ine-
quality, New York: McGraw Hill..
Kuznets, Simon (1955): Economic Growth and Inco-me Inequality, American Economic Review, 45:1-28.
Laumann, Edward O. (1966): Prestige and Association
in an Urban Community, Indianapolis: Bobbs-
Merrill.
Leigh, Andrew, Christopher Jencks and Timothy
Smeeding (2009): Health and Economic Inequa-lity in W. Salverda, B. Nolan and T. Smeeding(eds.), Oxford Handbook of Economic Inequality,
Oxford: Oxford University Press.
Lynch, John W., George Davey Smith, George A. Ka-
plan, and James S. House (2000): Income Ine-
7/29/2019 2012 Goldthorpe Sociologia de La Desigualdad
15/15
Reis 137 enero marzo 2012 pp 201 216
John H. Goldthorpe 215
quality and Mortality: Importance to Health of
Individual Income, Pyschosocial Environment or
Material Conditions, British Medical Journal,320: 1200-1204.
Lynch, John et. al (2004): Is Income Inequality a De-
terminant of Population Health? (Parts 1 and 2),Milbank Quarterly, 82: 5-99 and 355-400.
McGovern, Patrick, Stephen Hill, Colin Mills and Mi-
chael White (2007): Market, Class, and Employ-
ment, Oxford: Oxford University Press.
Marmot, Michael (2004): Status Syndrome, Londres:
Bloomsbury.
Mellor, Jennifer M. and Jeffrey Milyo (2001): Re-exa-mining the Evidence of an Ecological Association
between Income Inequality and Health, Journal
of Health Politics, Policy and Law, 26: 487-522.
OCDE (2010): Economic Policy Reforms: Going for
Growth, Pars: OCDE.
Okun, Arthur M. (1975): Equality and Efficiency: the Big
Tradeoff, Washington D.C.: Brookings Institution.
Parsons, Talcott (1967): Sociological Theory and Mo-
dern Society, New York: Free Press.
(1971): The System of Modern Societies, En-glewood Cliffs: Prentice-Hall.
Rose, David and Eric Harrison eds. (2009):Social Class
in Europe: An Introduction to the European Socio-
Economic Classification, London: Routledge.
Torssander, Jenny and Robert Erikson (2009): Mari-tal Partner and Mortality: the Effects of the Social
Positions of Both Spouses,Journal of Epidemio-logy and Community Health, 63: 992-998.
and (2010): Stratification and Mortality - AComparison of Education, Class, Status and In-
come, European Sociological Review, 26: 465-474.
Treiman, Donald J. (1970): Industrialization and So-cial Stratification in E. O. Laumann (ed.), SocialStratification: Research and Theory for the 1970s,
Indianapolis: Bobbs-Merrill.
Weber, Max (1922-1968): Economy and Society, Ber-
keley and Los Angeles: University of California
Press.
Wilkinson, Richard G. (1996): Unhealthy Societies,
London: Routledge.
(2005): The Impact of Inequality, London: Rout-ledge.
and Kate E. Pickett (2009): The Spirit Level, Lon-don: Allen Lane.
RECEPTION: 04/02/2011
ACCEPTANCE: 08/04/2011